MCP
Connect AI assistants to your workspace to inspect recordings, captures, logs, and network data.
MCP (Model Context Protocol) lets you connect AI assistants — such as Claude Desktop and Cursor — directly to your Screendesk workspace. Once connected, the AI assistant can look up recordings, read bug reports, review console logs, and examine system information on your behalf. This means you can ask your AI assistant questions like "What error is the customer seeing in this recording?" or "Summarize the last 5 bug reports" and get answers grounded in your actual Screendesk data.
Plan Availability: Pro and Enterprise only
What AI Assistants Can Access
When you connect an AI assistant via MCP, it can:
List and read your recordings — titles, descriptions, transcripts, browser and OS details, screen dimensions, and duration
Read console logs — browser errors and warnings captured during a recording session
Read network requests — HTTP calls and responses logged during a recording
List and read your bug reports — titles, descriptions, screenshots, reporter information, labels, and assignees
The AI assistant sees exactly what you see in Screendesk — no more, no less. This is called permission mirroring: the token you generate carries your personal access level, so the AI can only reach recordings and bug reports that are already visible to you. If you are a Member who only sees recordings in certain folders, the AI has the same restriction.
Setting Up MCP
Copy the token
The token is displayed on the next screen along with a copy button. Copy it immediately and store it somewhere secure — you will not be able to see it again.
The page also shows setup instructions with ready-to-use configuration snippets for Claude Desktop and Cursor, with your token already filled in.
Configure your AI assistant
Paste the configuration into your AI assistant's settings. Below are the setups for the most common tools.
Add the following to your Claude Desktop configuration file:
{
"mcpServers": {
"screendesk": {
"url": "https://app.screendesk.io/mcp",
"headers": {
"Authorization": "Bearer YOUR_TOKEN_HERE"
}
}
}
}Replace YOUR_TOKEN_HERE with the token you copied.
Add the following to your Cursor MCP configuration:
{
"mcpServers": {
"screendesk": {
"url": "https://app.screendesk.io/mcp",
"headers": {
"Authorization": "Bearer YOUR_TOKEN_HERE"
}
}
}
}Replace YOUR_TOKEN_HERE with the token you copied.
Managing Your Tokens
Your active tokens are listed on the Profile → MCP Access page. For each token you can see:
A preview of the token (first few characters, masked)
When it was last used
When it was created
A Revoke button to permanently disable it
You can generate multiple tokens — for example, one for Claude Desktop and one for Cursor — and revoke them independently.
If you suspect a token has been compromised, revoke it immediately from your MCP Access settings. Revoking a token takes effect instantly and the AI assistant will lose access.
Token Expiration
MCP tokens expire 90 days after creation. When a token expires, the AI assistant will no longer be able to access your Screendesk data. You will need to generate a new token and update your AI assistant's configuration.
Permission Mirroring
MCP tokens inherit your personal permissions. This means:
Admin
All recordings and bug reports across the workspace
Editor
All recordings and bug reports across the workspace
Member
Only recordings and bug reports you have access to (based on folder membership, assignments, and mentions)
Watch-only
MCP access is not available — tokens cannot be created
If you belong to multiple workspaces, the AI assistant can access data from all of them, following the same permissions you have in each.
The AI assistant never gets more access than you have. If a recording is in a private folder you cannot see, the AI assistant cannot see it either.
Example Use Cases
Here are some ways support teams use MCP with their AI assistants:
Investigating a customer issue — "Look up recording abc123 and tell me what console errors were logged." The AI reads the recording's console logs and gives you a summary of the errors, saving you from scrolling through log output.
Preparing for a follow-up — "List the last 10 bug reports and highlight any that mention checkout failures." The AI scans recent captures and filters them by your criteria.
Getting context before a call — "Summarize the bug report with title 'Login page broken on Safari'." The AI pulls the report details, system information, and any console errors to give you a quick briefing.
Triaging incoming recordings — "What are the most recent recordings and do any of them show JavaScript errors?" The AI lists recordings and checks their console logs for you.
Frequently Asked Questions
Do I need to be an admin to set up MCP?
No. Any user who is not Watch-only can generate MCP tokens from their own Profile settings. The token reflects your personal permissions, so you do not need admin access to use MCP.
Can I use the same token in multiple AI assistants?
Yes — a single token can be used in multiple tools. However, for better security, we recommend generating a separate token for each application so you can revoke them independently.
What happens if I revoke a token?
The AI assistant immediately loses access to your Screendesk data. You can generate a new token at any time if you want to reconnect.
Can the AI assistant make changes to my recordings?
It can update titles, summaries, and descriptions on recordings and bug reports. It cannot delete recordings, change sharing settings, or modify workspace configuration.
Last updated
Was this helpful?